Remove duplicate column pairs, sort rows based on 2 columns [duplicate]

Answer 1

In case people are interested in solving this using dplyr:

library(dplyr)
testdata %>% 
   rowwise() %>%
   mutate(key = paste(sort(c(Var1, Var2)), collapse="")) %>%
   distinct(key, .keep_all=T) %>%
   select(-key)

# Source: local data frame [3 x 4]
# Groups: <by row>
# 
# # A tibble: 3 × 4
#    Var1  Var2       f       g
#   <dbl> <dbl>  <fctr>  <fctr>
# 1     1     4    blue    blue
# 2     2     3   green   green
# 3     5     5 orange2 orange1

Answer 2

If the data are large, as in Sort large amout of data and save repeated pairs of values in R, using apply() on each row will be expensive. Instead, create the set of unique values

uid = unique(unlist(testdata[c("Var1", "Var2")], use.names=FALSE))

determine whether a swap is necessary

swap = match(testdata[["Var1"]], uid) > match(testdata[["Var2"]], uid)

and update

tmp = testdata[swap, "Var1"]
testdata[swap, "Var1"] = testdata[swap, "Var2"]
testdata[swap, "Var2"] = tmp

remove duplicates as before

testdata[!duplicated(testdata[1:2]),]

If there were many additional columns, and copying these were expensive, a more self-contained solution would be

uid = unique(unlist(testdata[c("Var1", "Var2")], use.names=FALSE))
swap = match(testdata[["Var1"]], uid) > match(testdata[["Var2"]], uid)
idx = !duplicated(data.frame(
    V1 = ifelse(swap, testdata[["Var2"]], testdata[["Var1"]]),
    V2 = ifelse(swap, testdata[["Var1"]], testdata[["Var2"]])))
testdata[idx, , drop=FALSE]

Answer 3

Instead of sorting for the whole dataset, sort the 'Var1', 'Var2', and then use duplicated to remove the duplicate rows

testdata[1:2] <- t( apply(testdata[1:2], 1, sort) )
testdata[!duplicated(testdata[1:2]),]
#   Var1 Var2       f       g
#1    1    4    blue    blue
#2    2    3   green   green
#5    5    5 orange2 orange1